Physically-Based Modelling of co-seismic Landslide, Debris Flow and Flood Cascade
- 1Faculty of Geo-Information Science and Earth Observation (ITC), University of Twente
- 2Institute of Mountain Hazards and Environment, Chinese Academy of Sciences & Ministry of Water Conservancy
- 1Faculty of Geo-Information Science and Earth Observation (ITC), University of Twente
- 2Institute of Mountain Hazards and Environment, Chinese Academy of Sciences & Ministry of Water Conservancy
Abstract. The 2008 Wenchuan earthquake lead to various complex multi-hazard chains that included seismically-triggered landslide initiation, landslide run-out, river damming, dam breaching and flooding. The modelling of the interactions between such hazardous processes is challenging due to the complexity and uncertainty. Here we present an event-based physically-based model that is able to simulate multi-hazard land surface process chains within a single unified simulation. The final model is used to simulate a multi-hazard chain event in the Hongchun watershed, where co-seismic landslides led to a landslide dam and, two years later, a debris flow that breached the landslide dam. While most aspects of the multi-hazard chain are well predicted, the correct prediction of slope failures remains the biggest challenge. Although the results should be treated carefully, the development of such a model provides a significant progress in the applicability of multi-hazard chain simulations.
Bastian van den Bout et al.
Status: final response (author comments only)
-
RC1: 'Comment on nhess-2021-292', Martin Mergili, 21 Feb 2022
The authors present an application of an extended version of the OpenLISEM model for the integrated simulation of a landslide – debris flow – flood cascade related to the 2008 Wenchuan Earthquake, Sichuan, China. The process chain considered includes co-seismic landsliding (slope stability & runout) and rainfall-triggered post-seismic erosion of the landslide deposit, leading to a debris flow the deposit of which blocked the Min river, resulting in flooding of the Yingxiu town. The simulations are built on a diverse and detailed data acquisition campaign, needed to feed the data-hungry OpenLISEM model.
The work is highly relevant both from a scientific and from a practical point of view, and is certainly within the scope of the NHESS journal. Increasing our capacity of simulating complex landslide cascades is fundamental for better scenario-based predictions which can be used to inform risk management. This this work can be considered a major contribution in this direction, even though the uncertainties in the input data still represent an important limitation (which is addressed in the paper). Further, the manuscript is well written and illustrated, adequately describing the background, study area, data and methods, and results of the study. Challenges and limitations are adequately discussed.
As a consequence, I would certainly like to see this work published. Before recommending publication, I suggest some minor-moderate revisions. I have not identified any fundamental issues, but there are some rather minor suggestions from my side, including a number of language issues particularly towards the end of the manuscript.
Here are my detailed comments:
Eq. 21 and 22: Sx,f and Sy,f would be correct, I think.
236: Sf is the momentum source term for fluids, not for solids. Please thoroughly check all equations and variable descriptions for correctness.
Figure 4: It might be nice to add a photo showing the situation. I am sure the authors have plenty of field photos available – if not, I can offer the following: https://www.mergili.at/worldimages/picture.php?/7252/category/60
410: “pedotransfer functions” (not “pseudo transfer functions”), I think
440ff: Are the root systems deeper than the typical landslide depth, so that it is appropriate to consider root cohesion for slope stability?
455: r.slope.stability includes seismic forcing (Newmark and pseudostatic), but these functions were added very recently and have not yet been thoroughly evaluated.
Fig. 9: Nice figure, but maybe better put (D) as (A), this would seem more logical to me.
Table 6: Very promising results. However, one has to keep in mind that flow modelling is always easier than slope stability modelling, as the areas where the flow concentrates are prescribed by the topography.
561f: Please explain in some more detail the correspondence of modelled and observed timing (the text is not fully clear to me).
Fig. 12: Clearly indicate also in the legend what is observation and what is simulation.
589: “effects” instead of “effecta”
598: “as a result”
599f: “Despite its importance …”
604: entrainment constant or entrainment coefficient?
605ff: Note the new entrainment model of Pudasaini and Krautblatter (2021). [Pudasaini, S. P., Krautblatter, M. (2021). The mechanics of landslide mobility with erosion. Nature communications, 12(1), 1-15.]
5.2: This is a very important step. But shouldn’t it rather be described in the results instead of discussions?
Fig. 14: Nice figure, but (i) legends should show classes instead of values, and (ii) the colouring is not ideal, as the medium probabilities are much paler that the low probabilities, which gives a strange visual impression.
626: “show illustrate”: remove one of the two words, and: “Figure 14”, not “Figure 17”.
634: This statement is true in principle, but it has to be considered that the Huascarán events considered by Mergili et al. (2018b) were of a completely different type (extraordinarily rapid and energy-rich, with air-lifting of material, etc.), and there was the specific situation of the ridge that was overtopped, completely changing the impact and inducing threshold behaviour. I suggest to briefly mention that this comparison has to be interpreted with care.
644: Check language.
658: Mergili et al. 2018 a or b?
662f: In principle, this type of hazard chain is not so rare, but its magnitude and specific characteristics are.
680: “first sequence”
683: “on its banks”
688: “still under development” – is this wording appropriate when citing a 2011 paper?
693: “elevation models”
713: “Authors C. Tang and C. Tang”: there is only one author named C. Tang provided in the list of authors.
723: “A dataset”
-
AC1: 'Reply on RC1', Bastian van den Bout, 23 May 2022
Dear Reviewer,
We would like to express our gratitude for the work in providing constructive feedback to our manuscript. We appreciate the feedback and are convinced the quality of the manuscript will be improved as a result.
As a response to the some of the detailed comments:
All other comments will be followed in the revision of the manuscript.236: Sf is the momentum source term for fluids, not for solids. Please thoroughly check all equations and variable descriptions for correctness.
-Indeed, this should be changed according to the suggestion of the reviewer.
236: Sf is the momentum source term for fluids, not for solids. Please thoroughly check all equations and variable descriptions for correctness.
-This check will be carried out
Figure 4: It might be nice to add a photo showing the situation. I am sure the authors have plenty of field photos available – if not, I can offer the following: https://www.mergili.at/worldimages/picture.php?/7252/category/60
-We appreciate the offered image and might use it, the angle is very nice!
410: “pedotransfer functions” (not “pseudo transfer functions”), I think
-Indeed, this is referring to pedotransfer function.
440ff: Are the root systems deeper than the typical landslide depth, so that it is appropriate to consider root cohesion for slope stability?
-The depth of the root systems is still somewhat uncertain. It is likely based on the type of vegetation that it surpasses several meters. However, the co-seismic landslides might exceed this depth. We will clarify this in the manuscript. However, we implemented the vegetation effect predominantly for the modelling of erosion, where the additional cohesion provides increased resistance to entrainment.
561f: Please explain in some more detail the correspondence of modelled and observed timing (the text is not fully clear to me).
-We will clarify this in the revised text.
634: This statement is true in principle, but it has to be considered that the Huascarán events considered by Mergili et al. (2018b) were of a completely different type (extraordinarily rapid and energy-rich, with air-lifting of material, etc.), and there was the specific situation of the ridge that was overtopped, completely changing the impact and inducing threshold behaviour. I suggest to briefly mention that this comparison has to be interpreted with care.
-For sure, we will add this to the manuscript. We did not mean to compare the studies in the nature of the processes. Instead, we hoped to focus on threshold effects as a challenge for multi-hazard modelling and dealing with its intrinsic uncertainties.
-
AC1: 'Reply on RC1', Bastian van den Bout, 23 May 2022
-
RC2: 'Comment on nhess-2021-292', Tommaso Baggio, 07 Mar 2022
The authors presented a new version of the model OpenLISEM for which they introduced the implementation of different features involving the slope stability/failure and bed entrainment. They successively described the cascade process that lead to the flood of the town of Yinxiu in China, back calculating it.
The study resulted of high interest for the scope of the journal and it provided new advances in the topic of cascade process numerical simulations involving mass flow movements and related phenomena. The value of the study is increased by the use of physical based models together with an accurate preparation of the input data of which some of them were acquired on the field. Results and limitations of the study were well discussed. The paper is well structured and written in good English.
Consequently I would consider the publication of the work after moderate/minor revisions.
Main comment: It is not really clear if the main objective of the study is: (i) the presentation of a new version of the OpenLISEM model, (ii) the possibility to simulate the mass flow process cascade occurred in Yinxiu or (iii) both of them. In my opinion from the title it looks like the first case (i) while from the abstract the second one (ii). Try to clarify this aspect.
Minor comments:
65-66 Please add at least a reference supporting this statement. e.g. Baggio et al. (2021) simulated the process of debris flow initiation through bed erosion releasing an input hydrograph characterized by a solid concentration equal to the 10 % respect the total input volume. (Baggio, T., Mergili, M., & D'Agostino, V. (2021). Advances in the simulation of debris flow erosion: The case study of the Rio Gere (Italy) event of the 4th August 2017. Geomorphology, 381, 107664.)
99-100 It is not clear if the entire model is presented in the study for the first time or if in the study is presented an updated version of the model characterized by new important features.
339 Please specify the range of the terrain slope of the watershed instead of “> 30 degrees”. If possible provide information about the type of vegetation (dense forest, shrubs,…)
345 Is there an estimation of the total volume of the moved material?
Fig.4 To improve the understanding of the process you may consider to add a dem of difference map involving the pre- and post-earthquake DTMs.
351 satellite instead of “ssatellite”
356 Please provide also the value of the second highest peak in rainfall and the total duration of the storm event. Here you can directly refer to Figure 12 for the rainfall pattern and to Table 1 for the rainfall source.
358 Can you provide also the length and width of the debris dam?
Tab. 1 Please provide the acquisition dates of the DTMs and the position of the rainfall station in terms of distance to the watershed outlet.
385 The end of the sentence is not really clear. A suggestion could be “…, we resampled the input base maps to the final resolution of 10 meters.”
391 These inputs refer to simulation 1? Please specify it
499 I suppose the correct number of the figure is 9D instead of 12
528 Please provide the value of the simulated deposit volume producing the debris dam and make a comparison with the observed one if possible.
529 I suppose the correct figure number is 10.
Fig. 11 In the legend I would substitute “water depth” with “flow depth” since the model is biphasic.
548 regarding the Min river, what is the discharge value used to reproduce the flow of the river? Could you also provide the location of the input hydrograph in Figure 11?
Fig. 12 The scale of the rainfall intensity is not reported. Please also consider to use another colour for the rainfall intensity since the figure is not clear. Moreover, I noticed a constant value equal to 0 in correspondence of the first rainfall peak regarding the fluid height, while the fluid+solid height is around 3 meters. Is it correct?
626 I think it is Figure 14. Please check.
610 It would be interesting to calculate the total volume of simulated eroded material entrained downstream the dam breach and to successively compare it with the original dam-debris volume. Furthermore, consider also to derive the mean eroded volume for channel length (downstream the dam breach) in order to compare it with other studies involving debris flow erosion.
642 If available please report the estimated return period of the rainfall event to assess its magnitude.
691 “Were” is repeated two times.
-
AC2: 'Reply on RC2', Bastian van den Bout, 23 May 2022
Dear Reviewer,
We would like to express our gratitude for the work in providing constructive feedback to our manuscript. We appreciate the feedback and are convinced the quality of the manuscript will be improved as a result.
For the purpose of the manuscript: We primarily wanted to focus on option ii) the possibility to simulate the mass flow process cascade around Yinxiu. We will adapt the title if possible, and clarify further in the text of the revised manuscript.
As a response to the some of the detailed comments:
All other comments will be followed in the revision of the manuscript.65-66 Please add at least a reference supporting this statement. e.g. Baggio et al. (2021) simulated the process of debris flow initiation through bed erosion releasing an input hydrograph characterized by a solid concentration equal to the 10 % respect the total input volume. (Baggio, T., Mergili, M., & D'Agostino, V. (2021). Advances in the simulation of debris flow erosion: The case study of the Rio Gere (Italy) event of the 4th August 2017. Geomorphology, 381, 107664.)
- We will clarify this part with the references specified by the reviewer.
99-100 It is not clear if the entire model is presented in the study for the first time or if in the study is presented an updated version of the model characterized by new important features.
- The model is an updated version of the OpenLISEM model. More information on this model can be found on lisemmodel.com We will also clarify this in the text.
339 Please specify the range of the terrain slope of the watershed instead of “> 30 degrees”. If possible provide information about the type of vegetation (dense forest, shrubs,…)
- We will add this to the manuscript text.
345 Is there an estimation of the total volume of the moved material?
- We don’t know of any such estimate in the literature. This is a difficult quantity to estimate, even with the model in place, as erosion and deposition overlap spatially. Erosion might occur first with deposition later. Of course, the model can state total erosion, which could be calculated from the erosion map, but this can better be compared to observed elevation differences.
Fig.4 To improve the understanding of the process you may consider to add a dem of difference map involving the pre- and post-earthquake DTMs.
351 satellite instead of “ssatellite”
- We will adapt the text accordingly
356 Please provide also the value of the second highest peak in rainfall and the total duration of the storm event. Here you can directly refer to Figure 12 for the rainfall pattern and to Table 1 for the rainfall source.
358 Can you provide also the length and width of the debris dam?
- We will add these to the revised manuscript text.
Tab. 1 Please provide the acquisition dates of the DTMs and the position of the rainfall station in terms of distance to the watershed outlet.
385 The end of the sentence is not really clear. A suggestion could be “…, we resampled the input base maps to the final resolution of 10 meters.”
- We will adapt the text accordingly
391 These inputs refer to simulation 1? Please specify it
-they do, except for the NDVI, which was provided based on the appropriate moment in time, as the event took place over the duration of several years.
499 I suppose the correct number of the figure is 9D instead of 12
- We will adapt the text accordingly
528 Please provide the value of the simulated deposit volume producing the debris dam and make a comparison with the observed one if possible.
- We would like to refer to line 630-640 in the manuscript text.
529 I suppose the correct figure number is 10.
- We will adapt the text accordingly
Fig. 11 In the legend I would substitute “water depth” with “flow depth” since the model is biphasic.
- We will adapt the text accordingly
548 regarding the Min river, what is the discharge value used to reproduce the flow of the river? Could you also provide the location of the input hydrograph in Figure 11?
- The discharge was estimated to be at bankfull conditions based on local reports, thus a fixed-height boundary conditions was added to the Min-RIver for the height of the embankments. We will specify this in the text.
- For the location of the hydrograph, We will adapt the text accordingly
Fig. 12 The scale of the rainfall intensity is not reported. Please also consider to use another colour for the rainfall intensity since the figure is not clear. Moreover, I noticed a constant value equal to 0 in correspondence of the first rainfall peak regarding the fluid height, while the fluid+solid height is around 3 meters. Is it correct?
- We will adapt the text accordingly. The lack of solids during the first rainfall peak is correct. There is not yet sufficient shear stress on the available sediment with vegetation to induce erosion. The only erosion taking place is upstream, and doesn’t travel to the outlet yet. Only after the main rainfall peak does the erosion reach downstream, but more importantly, the sediment from upstream travels downstream.
626 I think it is Figure 14. Please check.
- We will adapt the text accordingly
610 It would be interesting to calculate the total volume of simulated eroded material entrained downstream the dam breach and to successively compare it with the original dam-debris volume. Furthermore, consider also to derive the mean eroded volume for channel length (downstream the dam breach) in order to compare it with other studies involving debris flow erosion.
642 If available please report the estimated return period of the rainfall event to assess its magnitude.
- We have heard unverified estimates from local hydrologists, and those responsible for the warning system in that area, but have not been able to find any data to back up these claims. We will check again if any information on this is available in the literature.
691 “Were” is repeated two times.
- We will adapt the text accordingly
-
AC2: 'Reply on RC2', Bastian van den Bout, 23 May 2022
Bastian van den Bout et al.
Bastian van den Bout et al.
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
231 | 161 | 12 | 404 | 6 | 7 |
- HTML: 231
- PDF: 161
- XML: 12
- Total: 404
- BibTeX: 6
- EndNote: 7
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1